Hacker News Reader

Comment by throitallaway | Hacker News Reader

When Super Mario 3D All-Stars for Switch came out I did alright on Super Mario 64 until I encountered a level that required precisely timed wall jumps in order to advance. I consistently missed my jumps and wondered what was up. I plugged in my controller via USB and was still failing to do the jumps properly. What finally fixed it was putting the input on my TV into "game" mode, which reduced the amount of processing/latency. If local display latency can cause issues with gaming, network latency would be a non-starter in a lot of cases.

View on HN

Those "old games" were designed around the technology of the time, which had drastically lower latency than anything today except for high end gaming (and sometimes not even that).

- Controller buttons caused CPU interrupts, so basically 0 latency

- No OS getting in the way

- N64 era would be double buffered and then straight to the CRT

- NES era would literally calculate the pixels in real time as the CRT beam moved across the screen

- CRTs have virtually no latency, same with the analog signal chain because there is no buffering

So when people try to play them in modern systems, things that were easy back then are quite hard now.

There's a reason anybody speedrunning SM64 will play on a CRT.

In 2015 I found myself a decent deal on my old childhood console, the Super Nintendo which I had sold at a garage sale years ago (and later regretted)

I bought a Japanese copy of Super Mario World on ebay (Japanese copies of games were peanuts at the time, I assume they're more now), and found an old CRT for cheap on a local used site. I continued to play SMW many times over the years on various platforms and emulators, and I'm pretty damn good at the game.

But man, did it blow my mind feeling as little input latency as I did the first time I booted it on a CRT after all those years. It actually took a little bit of time to adapt to. It's like that phenomenon where if a button activates a light with low enough latency, people think the light is predicting when they'll hit the button, i.e. turning on before the button is pressed. People don't realize the latency we started dealing with when everything went from analogue to digital!

>> "NES era would literally calculate the pixels in real time as the CRT beam moved across the screen"

Relevant: https://en.wikipedia.org/wiki/Raster_interrupt#Nintendo_Ente...

Surprisingly the display latency on TVs is regularly over 100ms, way more than latency at most homes these days (about 10ms for Comcast here).

That is amazingly inaccurate. If you count the entire chain from input to processing to display you'll be under 40-50ms.

Nah, it's definitely true. I have seen plenty of (especially larger-format) TVs where, with "game mode" (or the respective equivalent) disabled, it's unbearable to even do latency-forgiving tasks like office work on them.

It really depends on the TV. And if your gaming console is plugged into a receiver, it could add more.

My previous TV had about 150 ms of video latency. Even if I enabled Game Mode, it was 75 ms, which was still noticeable.

With my current TV, I have no idea what the latency is because I stopped gaming on console and so I'm not playing Rock Band which had a calibration option to compensate for video and audio latency.

I've had a TV that were nearly 200ms, but it was the absolute cheapest panel I could find.

If you don't configure your TV correctly sure, you could get massive amounts of lag. Even with an OLED display if you turn on all the post processing you're going to have problems. That's not really a fault of the TV though. I disable almost all post processing on my TVs and get a better picture without the downsides.

With digital receivers adding lag is minimal at best. Especially with newer models that don't draw on top of the source signal.

It is quite accurate, but only for the highly processed cinematic modes. Latency is essentially irrelevant for video content as long as the audio stays in sync. The game mode drops all the processing and is usually pretty low.

For example, my LG OLED tv is regarded as one of the best for gaming. Game mode latency is about 9ms. Cinema mode is about 90ms.

I only know for sure from the few displays I've looked at myself but this table from a quick Google search is consistent:

https://www.rtings.com/tv/tests/inputs/input-lag

Yeah, I mean if you enable post processing (which is most likely garbage) you're going to get bad input lag.

Yeah, the latency on a lot of TVs outside of "Game Mode" is really atrocious, commonly in the order of 100 - 200ms, which is way higher than even network latency on game streaming services assuming you're close to the data center.

Worth noting, I'm pretty sure by default when you plug a Switch controller into the dock, it just charges and continues to communicate over Bluetooth. Someone can correct me if I'm wrong

Super Mario 3D all star is running emulation on the switch. How much latency is introduced with just that? It was not like the emulation was even top notch either.

It's hard to say. If the emulation to compute a single frame finishes before the frame deadline elapses then the latency caused by emulation is effectively zero.

Reply on news.ycombinator.com